Re: pseudoinverse

Are the elements in S (diagonal) matrix sorted according to their
alphebatic order in your numerical recipes code? Is it possible that
the numerical recipes code scale S (and either U and V) differently ?
Or is it possible that it returns "V" instead of "transpose(V)" (some
software does this)?

These are just my guessings. I only have experience using matlab and
other fortran codes for svd, but not scilab and numerical receipe's
code. :(



On 2005-09-17, andrea mordini <stredostudio@xxxxxxxxx> wrote:
> What u write is correct but i've tried to compute the SVD
> with scilab and with the code provided by numerical recipes
> and i obtain different U S and V matrices.
> In both the cases A=U S transpose(V) but the calculation of the pinv is
> different in my code (wrong) and in scilab (correct)
> How is this possible?
>
> andrea
>
>
> SC Huang wrote:
>> OK, I am not sure if this is what you are looking for:
>>
>> If the svd of matrix is A is:
>> A = U S V^*
>> then the pseduoinverse of A is:
>> pseudo-inv(A) = V S^{-1} U^*
>>
>> Note that S is diagonal, so S^{-1} is easy to calculate. Also, V^*
>> denotes the hermitian of V (complex cojugate transpose). If V is real,
>> then V^* is just the transpose of V. Hope this helps.
>>
>>
>> On 2005-09-16, andrea mordini <stredostudio@xxxxxxxxx> wrote:
>>
>>>I'm searching the fortran code to compute the pseudoinverse matrix by
>>>the SVD single value decomposition which i have already computed.
>>>thanks
>>>
>>>andrea
.

Relevant Pages

  • Re: pseudoinverse
    ... andrea mordini wrote: ... > What u write is correct but i've tried to compute the SVD ... > with scilab and with the code provided by numerical recipes ...
    (comp.lang.fortran)
  • Re: linalg[leastsqrs] in Maple V R4
    ... For some reason I thought that linalgwould've used an SVD ... Numerical Recipes sec. 5.13 prescribe an SVD approach for the very kind ... I thought that the LinearAlgebra routine LeastSquares was more ...
    (sci.math.symbolic)
  • Re: LSA (latent sematic analysis) in Lisp, or at least SVD
    ... r> Why would you want a CL implementation of SVD? ... i'm now using Armed Bear Common Lisp that can easily FFI only Java. ... I cannot speak for SVD in particular, but "numerical recipes" is ... I am always happier getting my matrix factorizations from ...
    (comp.lang.lisp)
  • Re: pseudoinverse
    ... What u write is correct but i've tried to compute the SVD ... with scilab and with the code provided by numerical recipes ... In both the cases A=U S transposebut the calculation of the pinv is different in my code and in scilab (correct) ... the SVD single value decomposition which i have already computed. ...
    (comp.lang.fortran)
  • Re: pseudoinverse
    ... with scilab and with the code provided by numerical recipes ... In both the cases A=U S transposebut the calculation of the pinv is different in my code and in scilab ... All that is determined is the subspace and not its representation. ... That means that there can arbitrary orthogonal transformations permitted which means there can be many right answers, each corresponding to a different choice of the arbitrary transformation. ...
    (comp.lang.fortran)